Push to HF space

data/lilac.yml

@@ -228,6 +228,12 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: text
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -399,6 +405,12 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: text
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -562,6 +574,12 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: text
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -2836,6 +2854,355 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: legal-termination
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: negative-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: non-english
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: positive-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: profanity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: question
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: source-code
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-instruction
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: toxicity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-context
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path: original-response
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: legal-termination
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: negative-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: non-english
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: positive-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: profanity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: question
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: source-code
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path: original-response
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: toxicity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: legal-termination
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: negative-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: non-english
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: positive-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: profanity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: question
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: source-code
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-instruction
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: toxicity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-context
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: legal-termination
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: negative-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: non-english
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: positive-sentiment
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: profanity
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: question
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: source-code
+          version: null
+          draft: main
+          signal_name: concept_score
+      - path:
+          - new-response
+          - value
+          - '*'
+        signal:
+          embedding: gte-small
+          namespace: lilac
+          concept_name: toxicity
+          version: null
+          draft: main
+          signal_name: concept_score
     settings:
       ui:
         media_paths:
@@ -3415,6 +3782,18 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: question
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path: response
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -3807,6 +4186,22 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path:
+          - translation
+          - es
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path:
+          - translation
+          - en
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -3973,6 +4368,12 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: lecture
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -4136,6 +4537,12 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: text
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -4298,6 +4705,12 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path: overview
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths:
@@ -4609,6 +5022,28 @@ datasets:
           version: null
           draft: main
           signal_name: concept_score
+      - path:
+          - outline
+          - '*'
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path:
+          - concepts
+          - '*'
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
+      - path: markdown
+        signal:
+          embedding: gte-small
+          eps: 0.05
+          min_samples: 5
+          signal_name: cluster_dbscan
     settings:
       ui:
         media_paths: